Workflow Tutorials

This chapter walks you through ten complete, self-contained tutorials. Each tutorial builds a real workflow from scratch and shows exactly which nodes to use, how to wire them, and what to expect in the log panel. All examples use the built-in node set unless explicitly noted.

---

Tutorial 1: Basic File Renaming Workflow

Goal: Load a list of filenames from a folder, apply a naming convention (lowercase with underscores), and rename every file on disk. Print a summary when done.

Nodes Used

Step-by-Step Build

1. Create the entry Python Script node

Open the Node Library (left panel) and drag a Python Script node onto the canvas. This will be the entry point — it fires first when you click Run.

In the node's code editor paste:

import os

folder = inputs.get("data", "")
if not folder or not os.path.isdir(folder):
    result = []
else:
    result = [
        os.path.join(folder, f)
        for f in os.listdir(folder)
        if os.path.isfile(os.path.join(folder, f))
    ]

Set the folder path as the default value on the data port (in the node's parameter editor). Connect exec_out to the ForEach node.

2. ForEach node

Drag a ForEach node onto the canvas. Connect:

• result output → ForEach items input
• Python Script exec_out → ForEach exec_in

ForEach fires its loop_exec_out once per item, passing the current item on the current_item port.

3. String Lower node

Drag String Lower. Connect:

• ForEach current_item → String Lower string input

This converts the full path to lowercase. However, we only want to lowercase the filename, not the whole path. Use a Python Script node instead:

import os

full_path = inputs.get("data", "")
folder = os.path.dirname(full_path)
old_name = os.path.basename(full_path)

new_name = old_name.lower().replace(" ", "_")
new_path = os.path.join(folder, new_name)

result = {"old_path": full_path, "new_path": new_path, "new_name": new_name}

4. Rename Python Script node

Add another Python Script node. Wire the transform script's result → this node's data input:

import os

info = inputs.get("data", {})
old_path = info.get("old_path", "")
new_path = info.get("new_path", "")

if old_path and new_path and old_path != new_path:
    try:
        os.rename(old_path, new_path)
        result = f"Renamed: {old_path} -> {new_path}"
    except OSError as e:
        result = f"ERROR: {e}"
else:
    result = f"Skipped (no change): {old_path}"

5. Console Print

Connect the result output to a Console Print node. Wire the rename node's exec_out back to the ForEach loop_exec_in to continue the loop.

6. Final summary

Connect ForEach exec_out (fires after the loop completes) to a Console Print with the message "Renaming complete.".

Running the Workflow

Set the folder path as the default value on the entry node's data port (or wire it from an upstream node). Click Run. Watch the log panel — each rename or skip appears in sequence.

---

Tutorial 2: Data Transformation Pipeline

Goal: Read a JSON file containing records (e.g., a list of people with names and birthdates), transform the data (uppercase names, reformat dates), and write the result to a new JSON file. Log a summary.

Nodes Used

Step-by-Step Build

1. File Load

Drag File Load onto the canvas. Set file_path to your source JSON file (e.g., people.json). The content output carries the raw text.

2. JSON Parse

Connect File Load.content → JSON Parse.json_string. The data output is the decoded Python object (list or dict).

3. Transform Python Script

import json
from datetime import datetime

records = inputs.get("data", [])
transformed = []

for record in records:
    new_record = dict(record)

    # Uppercase the name field
    if "name" in new_record:
        new_record["name"] = str(new_record["name"]).upper()

    # Reformat birthdate from YYYY-MM-DD to DD/MM/YYYY
    if "birthdate" in new_record:
        try:
            dt = datetime.strptime(new_record["birthdate"], "%Y-%m-%d")
            new_record["birthdate"] = dt.strftime("%d/%m/%Y")
        except ValueError:
            pass  # leave malformed dates untouched

    transformed.append(new_record)

result = json.dumps(transformed, indent=2)

4. File Save

Connect the transform script's result → File Save.content. Set file_path to people_transformed.json. Wire exec_out chain through.

5. Console Print

Add another Python Script node. Wire the transform script's result → its data input to build a summary:

import json
json_str = inputs.get("data", "[]")
try:
    count = len(json.loads(json_str))
except Exception:
    count = 0
result = f"Transformation complete. {count} records processed."

Then connect result → Console Print.

Key Points

• JSON Parse handles both list-of-dicts and plain dicts.
• All transform logic lives in one Python Script node, making it easy to add or remove transformation steps.
• File Save overwrites the target file. To write alongside the original, compute the output path dynamically.

---

Tutorial 3: Conditional Execution (If/Else Logic)

Goal: Check whether a numeric value exceeds a threshold. Route execution to an "above threshold" branch or a "below threshold" branch. Both branches converge at a final log node.

Nodes Used

Step-by-Step Build

1. Source value

# Simulated: in practice this could come from a file, API, etc.
result = {"value": 87.3, "threshold": 75.0}

2. Condition evaluator

Wire the source script's result → this node's data input:

data = inputs.get("data", {})
value = data.get("value", 0.0)
threshold = data.get("threshold", 0.0)
result = value > threshold

3. TwoWaySwitch

Connect:

• result (bool) → TwoWaySwitch condition input
• Condition evaluator exec_out → TwoWaySwitch exec_in

TwoWaySwitch has two exec outputs:

• exec_true — fires when condition is True
• exec_false — fires when condition is False

4. True branch

Connect exec_true → Console Print A. Set A's message port to a static string "Value exceeds threshold — triggering alert." or pipe a dynamic string through.

5. False branch

Connect exec_false → Console Print B with "Value is within threshold — no action needed.".

6. Convergence

Both Console Print nodes' exec_out pins connect to a shared final Console Print node with the message "Check complete.".

Note: In Vibrante-Node, multiple exec_out connections into a single exec_in are valid. The engine executes the target node once for whichever upstream exec fires. Since TwoWaySwitch fires exactly one branch, the final node runs exactly once.

Extending to Multi-Branch Logic

For more than two conditions, chain TwoWaySwitches:

Condition A → TwoWaySwitch 1
  exec_true  → branch A
  exec_false → Condition B → TwoWaySwitch 2
                 exec_true  → branch B
                 exec_false → branch C (default)

---

Tutorial 4: Loop Over a List

Goal: Take a list of names, process each one (title-case), collect the results, and print the final processed list.

Nodes Used

Step-by-Step Build

1. Create List

Use Create List (or a Python Script) to produce your list:

result = ["alice smith", "BOB JONES", "carol WHITE", "david brown"]

2. Initialize accumulator

Drag SetVariable. Set var_name = "processed_names", value = [] (empty list). Connect exec_out → ForEach exec_in.

3. ForEach

Connect result → ForEach items.

ForEach exposes:

• current_item — the value at the current index
• current_index — the 0-based index
• loop_exec_out — exec pin that fires for each iteration (connect your loop body here)
• exec_out — fires once after the loop finishes

4. Process item

Wire ForEach current_item → this script's data input:

item = inputs.get("data", "")
result = item.title()

5. GetVariable → List Append → SetVariable

Chain:

1. GetVariable: var_name = "processed_names" → current_list
2. List Append: list = current_list, item = result (from process script) → appended_list
3. SetVariable: var_name = "processed_names", value = appended_list

Wire the loop exec chain: ForEach loop_exec_out → Process Script → GetVariable → List Append → SetVariable → ForEach loop_exec_in (to continue the loop).

6. Post-loop

After ForEach exec_out:

1. GetVariable: "processed_names" → final_list
2. Python Script (wire GetVariable final_list → data): result = str(inputs.get("data", []))
3. Console Print

Expected Output

['Alice Smith', 'Bob Jones', 'Carol White', 'David Brown']

---

Tutorial 5: Maya Automation Workflow

Goal: Open a Maya scene headlessly, configure render settings, and export an Alembic cache. Check success or failure.

Nodes Used

Understanding the Action-List Pattern

Maya automation in Vibrante-Node uses an action list — a Python list of dicts that describes operations to perform. Each action node appends a dict to the list. The Maya Headless Executor node consumes the complete list, launches a Maya batch subprocess, and executes each action in order.

This design means:

• No Maya process is opened until the entire list is assembled.
• The action list can be branched, filtered, or modified before execution.
• Execution happens in a single subprocess — fast startup, no repeated Maya launch overhead.

Step-by-Step Build

1. Maya Action: Open Scene

This is a Houdini-style headless action node. If a custom maya_action_open_scene node is registered:

# Node python_code equivalent
actions = list(inputs.get("actions_in") or [])
actions.append({
    "type": "open_scene",
    "scene_path": inputs.get("scene_path", "")
})
return {"actions_out": actions, "exec_out": True}

Set scene_path to your .mb file path.

2. Maya Action: Set Render Settings

actions = list(inputs.get("actions_in") or [])
actions.append({
    "type": "set_render_settings",
    "renderer": inputs.get("renderer", "arnold"),
    "start_frame": int(inputs.get("start_frame", 1)),
    "end_frame": int(inputs.get("end_frame", 100)),
    "image_output": inputs.get("image_output", "/tmp/render/"),
    "image_prefix": inputs.get("image_prefix", "render")
})
return {"actions_out": actions, "exec_out": True}

3. Maya Action: Export Alembic

actions = list(inputs.get("actions_in") or [])
actions.append({
    "type": "export_alembic",
    "objects": inputs.get("objects", ""),       # e.g. "|geo_GRP"
    "output_path": inputs.get("output_path", "/tmp/export.abc"),
    "frame_range": [
        int(inputs.get("start_frame", 1)),
        int(inputs.get("end_frame", 100))
    ]
})
return {"actions_out": actions, "exec_out": True}

4. Maya Headless Executor

This node launches Maya, runs the actions, and outputs:

• success (bool)
• log_output (string)
• error_message (string, empty on success)

5. Conditional routing

Connect success → TwoWaySwitch condition. Route:

• exec_true → Console Print: "Maya export completed successfully."
• exec_false → Python Script that reads error_message and prints it.

Notes

• Ensure MAYA_LOCATION and VIBRANTE_MAYA_EXE are set in your environment.
• The Maya runner script in plugins/maya/ must have a handler for every action type key.
• objects can be a single DAG path or a semicolon-separated list.

---

Tutorial 6: Houdini Procedural Workflow

Goal: Create a Houdini geometry container, add a Box SOP, transform it, cook the network, and store the resulting node path.

Nodes Used

Step-by-Step Build

All Houdini nodes call the JSON-RPC bridge internally. The following Python Script node demonstrates the full flow in a single node (which you can then break into individual nodes as needed):

from src.utils.hou_bridge import get_bridge

async def execute(self, inputs):
    bridge = get_bridge()

    # 1. Ping to verify connection
    ping = bridge.ping()
    if ping.get("status") != "ok":
        self.log_error("Houdini not responding.")
        return {"result_path": "", "exec_out": True}

    # 2. Create /obj-level geo container
    geo_result = bridge.create_node("/obj", "geo", "tutorial_geo")
    geo_path = geo_result["path"]   # e.g. "/obj/tutorial_geo"

    # 3. Clear Houdini's default child nodes
    for child in bridge.children(geo_path):
        bridge.delete_node(child["path"])

    # 4. Create Box SOP
    box_result = bridge.create_node(geo_path, "box", "my_box")
    box_path = box_result["path"]

    # 5. Set box size
    bridge.set_parms(box_path, {"sizex": 2.0, "sizey": 1.0, "sizez": 1.5})

    # 6. Create Transform SOP
    xform_result = bridge.create_node(geo_path, "xform", "my_xform")
    xform_path = xform_result["path"]

    # 7. Wire box → xform
    bridge.connect_nodes(box_path, xform_path, output=0, input_idx=0)

    # 8. Translate the transform
    bridge.set_parms(xform_path, {"tx": 3.0, "ty": 1.0, "tz": 0.0})

    # 9. Set display and render flags on xform
    bridge.set_display_flag(xform_path, True)
    bridge.set_render_flag(xform_path, True)

    # 10. Cook
    bridge.cook_node(xform_path, force=True)

    # 11. Auto-layout
    bridge.layout_children(geo_path)

    self.log_info(f"Geo network created at: {geo_path}")
    return {"result_path": geo_path, "exec_out": True}

Expected Result

In Houdini's Network Editor, /obj/tutorial_geo appears with two SOPs (my_box → my_xform) wired in series, my_xform flagged for display and render, and the box visible in the viewport.

Extending to Export

After cooking, add:

# Save the hip file with the new network
save_result = bridge.save_hip("/path/to/output.hip")
self.log_info(f"Saved: {save_result['saved']}")

---

Tutorial 7: Prism Pipeline Asset Workflow

Goal: Initialize PrismCore, retrieve all assets for a project, get the latest export path for each asset, and log the results.

Nodes Used

Step-by-Step Build

Important: Place prism_core_init anywhere in the graph. The engine detects it before execution begins and bootstraps PrismCore automatically. You do not need to wire core between nodes.

1. prism_core_init

Drag onto canvas. Configure project_path if your Prism installation requires it. This node has no meaningful output — it exists solely for the bootstrap side effect.

2. prism_get_assets

from src.nodes.base import BaseNode
from src.utils.prism_core import resolve_prism_core

class Prism_Get_Assets(BaseNode):
    name = "prism_get_assets"

    async def execute(self, inputs):
        core = resolve_prism_core(inputs)
        if core is None:
            self.log_error("PrismCore not available.")
            return {"assets": [], "exec_out": True}
        try:
            assets = core.getAssets()
            return {"assets": assets, "exec_out": True}
        except Exception as e:
            self.log_error(f"Prism error: {e}")
            return {"assets": [], "exec_out": True}

3. ForEach → prism_get_export_path

For each asset dict, call:

core = resolve_prism_core(inputs)
asset = inputs.get("current_item", {})
asset_name = asset.get("asset", "")

try:
    paths = core.getExportPaths(asset=asset_name)
    latest = paths[-1] if paths else None
    return {"export_path": latest, "asset_name": asset_name, "exec_out": True}
except Exception as e:
    self.log_error(f"Could not get export path for {asset_name}: {e}")
    return {"export_path": None, "asset_name": asset_name, "exec_out": True}

4. Console Print

Build message (custom node receiving asset_name and export_path as named inputs):

async def execute(self, inputs):
    name = inputs.get("asset_name", "")
    path = inputs.get("export_path", None)
    return {"message": f"  {name}: {path or 'No export found'}", "exec_out": True}

Expected Output

character_hero: /project/exports/assets/character_hero/model/v003/character_hero.abc
prop_sword:     /project/exports/assets/prop_sword/model/v001/prop_sword.abc
environment_castle: No export found

---

Tutorial 8: Multi-Shot Rendering Workflow

Goal: Iterate over a list of shots. For each shot, open the corresponding Houdini scene, set the frame range, trigger a render, and collect the output paths.

Nodes Used

Shot List Builder

outputs["shots"] = [
    {"name": "sh010", "start": 1001, "end": 1120, "hip": "/project/scenes/sh010.hip"},
    {"name": "sh020", "start": 1001, "end": 1085, "hip": "/project/scenes/sh020.hip"},
    {"name": "sh030", "start": 1001, "end": 1240, "hip": "/project/scenes/sh030.hip"},
]

Per-Shot Processing Script

from src.utils.hou_bridge import get_bridge

async def execute(self, inputs):
    shot = inputs.get("current_item", {})
    bridge = get_bridge()

    hip_path = shot.get("hip", "")
    start = shot.get("start", 1)
    end = shot.get("end", 240)
    name = shot.get("name", "unknown")

    try:
        # Load the hip file
        bridge.run_code(f"hou.hipFile.load('{hip_path}', suppress_save_prompt=True)")

        # Set frame range
        bridge.set_playback_range(start, end)

        # Cook the mantra/karma output driver
        out_path = f"/out/mantra_beauty"
        bridge.cook_node(out_path, force=True)

        self.log_info(f"Rendered {name}: frames {start}-{end}")
        return {"render_result": {"shot": name, "status": "ok", "frames": end - start + 1}, "exec_out": True}

    except Exception as e:
        self.log_error(f"Failed to render {name}: {e}")
        return {"render_result": {"shot": name, "status": "error", "error": str(e)}, "exec_out": True}

Post-Loop Summary

results = inputs.get("data", [])
ok = [r for r in results if r.get("status") == "ok"]
err = [r for r in results if r.get("status") == "error"]

lines = [f"Render complete: {len(ok)} succeeded, {len(err)} failed."]
for r in err:
    lines.append(f"  FAILED: {r['shot']} — {r.get('error', '')}")

result = "\n".join(lines)

---

Tutorial 9: GroupNode (Subgraph) Workflow

Goal: Build a reusable "normalize filename" subgraph, collapse it into a GroupNode, and use it in two different workflows without duplicating nodes.

Step 1: Build the Reusable Logic

Create three nodes on the canvas:

Trim & Lower:

s = inputs.get("data", "")  # GroupIn value wired to data port
result = s.strip().lower()

Replace Spaces:

s = inputs.get("data", "")  # wired from previous script's result
result = s.replace(" ", "_").replace("-", "_")

Remove Special Chars:

import re
s = inputs.get("data", "")  # wired from previous script's result
result = re.sub(r"[^a-z0-9_.]", "", s)

Wire them in sequence: exec_out → exec_in, and data outputs → inputs.

Step 2: Add GroupIn and GroupOut Nodes

Drag a GroupIn node. Set its port_name parameter to "raw_name". This exposes an input port on the eventual GroupNode.

Drag a GroupOut node. Set its port_name to "final_name". Wire the Remove Special Chars result output → GroupOut value input.

Wire exec: GroupIn → Trim & Lower → Replace Spaces → Remove Special Chars → GroupOut.

Step 3: Collapse into a GroupNode

Select all five nodes (Ctrl+A or drag-select). Press Ctrl+Shift+G (or Edit → Group Selection).

A dialog prompts for a group name. Enter "normalize_filename".

The five nodes collapse into a single GroupNode with:

• Input port: raw_name
• Output port: final_name
• Exec pins: exec_in, exec_out

Step 4: Use the GroupNode

The GroupNode behaves identically to any other node:

[Python Script: generate raw name]
      |  (raw_name)
      v
[GroupNode: normalize_filename]
      |  (final_name)
      v
[Console Print]

Step 5: Inspect the Subgraph

Double-click the GroupNode to open its subgraph in a new tab. The tab is labeled [normalize_filename] and is fully editable — changes sync back to the parent workflow automatically.

Step 6: Reuse in Another Workflow

Save the workflow containing the GroupNode. In a new workflow, use File → Import Nodes (or copy-paste the GroupNode widget). The entire subgraph is embedded — no external dependencies.

---

Tutorial 10: Debugging a Failing Workflow

Goal: A workflow is producing incorrect output. Use Vibrante-Node's debugging tools to find and fix the problem.

Tool 1: The Log Panel

The log panel (bottom of the window) shows:

• Node started / finished with timing: Node 'Get Asset' finished in 0.34s
• Info messages from self.log_info()
• Error messages from self.log_error() in red
• Execution order — nodes appear in the order they ran

Start here. Look for red error lines. Click a log entry to highlight the corresponding node on the canvas.

Tool 2: Console Print Nodes

Insert a Console Print node between any two nodes to inspect the intermediate value:

[Source Node] → (data output) → [Console Print] → (passthrough) → [Downstream Node]

Console Print has a passthrough so it does not break the exec chain. This is the fastest way to verify that a value is what you expect at each stage.

Tool 3: Wire Hover Tooltips (Live Value Inspector)

After running the workflow, hover your mouse over any wire (edge) connecting two nodes. A tooltip appears showing:

port_name: <repr of last value, capped at 300 chars>

This works for both data wires and exec wires. No extra nodes needed — just hover.

Values persist after execution finishes, so you can inspect the entire graph's state at leisure.

Tool 4: Bypass a Node

Right-click any node → Bypass. A bypassed node is skipped during execution; its input data is forwarded directly to its output. This lets you disable a section without deleting it.

Use bypass to isolate which node is causing a problem:

1. Bypass the suspected node → run → does the error disappear?
2. If yes: the problem is in that node's logic.
3. If no: bypass the next node in the chain and repeat.

Common Failure Patterns

Step-by-Step Debug Session

1. Run the workflow. Note which nodes appear in the log.
2. If execution stops early, the last node in the log is the culprit or its immediate successor.
3. Insert Console Print nodes before and after the suspect node.
4. Re-run. Compare actual vs. expected values in the log.
5. Hover wires around the suspect node to see raw values.
6. Bypass the suspect node to confirm it is the source.
7. Open the node's code editor and add print() statements for more detail.
8. Fix the logic, remove debug Console Print nodes, re-run to confirm.